Risk-Based GST Refunds: When Provisional Refunds Become Discretionary
- Mayank Khichar
- Mar 27
- 10 min read
[Mayank is a student at NALSAR University of Law.]
The goods and services tax (GST) is conceptually designed as a destination-based consumption tax that attempts to mitigate the “tax on tax” effect (i.e. cascading taxes) through input tax credit (ITC). The ITC can be claimed on the value addition at each stage of the supply chain, and can be deducted by the assessee at time of filing the tax return. Simply put, under the ITC framework there is no need to refund post facto as the assessee may deduct the tax already paid on inputs at the time of filing of taxes. However, under the inverted duty structure (IDS) tax structure, envisaged under Section 54(3) CGST Act, 2017 read with Rule 89(5) Central Goods and Services Tax Rules 2017 (Rules), the tax rate on inputs (raw materials) is higher than the tax rate on the output (finished goods). This structure consequently requires the balance tax credit amount to be refunded.
Accordingly, the Central Board of Indirect Taxes and Customs (CBIC) has introduced a risk‑based mechanism for the grant of 90% provisional GST refunds, which are contingent on a system‑generated “risk” classification. This is given the ‘fact of law’ that, neither the Central Goods and Services Tax Act 2017 (Act) nor the Rules defines who an “low‑risk” or “non‑low‑risk” / “high-risk” applicant is. Consequently, the timing of refund depends not merely on compliance with Section 54, but on undisclosed algorithmic assessments, that effectively determine whether refundable ITC remains blocked during detailed scrutiny.
Accordingly, in this analysis, the author argues that (i) even after the amendment to Rule 91(2), the provisional‑refund regime brought in through a executive ‘notification’ is ultra vires the Act, and (ii) the CBIC Instruction No. 06/2025-GST (impugned arrangement / Instruction) suffers from the vice of ‘excessive delegation’ and the tax administration has turned legitimate ‘provisional refund’ into a privilege, controlled by undisclosed risk scores, while it ought be a norm given the explicit provision i.e. Section 54 of the Act and, (iii) the prolonged retention of refundable ITC violates the right to carry on trade under Article 19(1)(g) of the Constitution of India due to unreasonable blocking-cum-retention of working capital., that should ideally be freely available with the applicant-assessee for ‘trade’.
What is the Statutory Scheme on IDS and ITC?
Under Section 54(3) of the Act, which deal with ‘refund of tax’ provides that a registered applicant may claim refund of any unutilised ITC at the end of any tax period, when such credit has accumulated on account of rate of tax on inputs being higher than the rate of tax on output supplies. This provision also admits statutory exclusions where the assessee may not be eligible for refund, in case of specific notification ‘by the Government on the recommendations of the Council’. Notably, there is no exception vis-à-vis an assessee who may be classified as ‘high-risk’.
Further, Rule 89(5) of the Rules provides for the formula for calculation of ‘refund on account of inverted duty structure’ and Rule 91(2), reproduced below, states that:
Rule 91. Grant of provisional refund.-
[(2) The proper officer, on the basis of identification and evaluation of risk by the system, shall make an order in Form GST RFD-04, within a period not exceeding seven days from the date of the acknowledgement under sub-rule (1) or sub-rule (2) of Rule 90:
Provided that the proper officer, for reasons to be recorded in writing, may not grant refund on provisional basis and proceed with the order under Rule 92.
Here, one may note that the concerned officer in his independent judgement is required to decide on the ‘grant of refund’, which is to be based on ‘evaluation of risk by the system’. However, both the Rules and the Act is silent on (a) what constitutes “risk”, (b) the parameters or weights used, (c) the data inputs relied upon, and (d) the threshold for “high-risk” classification.
What Follows Such Classification?
In impugned arrangement, notably, there exists a gap between the ‘legal recognition’ of refund as an enforceable credit right under Section 54(3), as held by the Supreme Court (SC) in Union of India v. VKC Footsteps India (Private) Limited (VKC) and its ‘practical negation’ through delayed processing and restrictive (risk-based) refund criteria. This is especially because the accumulated ITC under an inverted duty structure remains unusable and is further delayed by the denial of provisional refunds during extended scrutiny of eligible but purportedly ‘high-risk’ applicants of refund of tax.
This delay in refunding the rightful money has immediate and tangible economic consequences for businesses, particularly manufacturers, MSMEs, and exporters. When refundable credit remains blocked for prolonged periods, firms face a working‑capital crunch, forcing them to rely on short‑term borrowing to fund routine operations such as procurement, wages, and compliance costs, thereby increasing interest burdens and reducing profitability weakening competitiveness, especially for exporters operating on thin margins in global markets. Over time, persistent refund delays also undermine supply‑chain stability, incentivise informal credit arrangements, and erode confidence in GST’s promise of tax neutrality. This arguably converts what is meant to be a consumption tax into a cost borne by businesses, rather than a mere pass‑through to consumers.
In this context, it might be beneficial to peruse the specific observations of the court in VKC, where the court noted that:
"The proviso to Section 54(3) merely prescribes the condition and does not deal with the quantum of refund since the quantum is prescribed by the substantive provision… Thus, the main provision of Section 54(3) confers an entitlement to the refund of the entire unutilised ITC and the proviso only seeks to provide the condition and not to obliterate the main provision."
Analysis
The arrangement is ultra vires the Act
At the threshold, one might note that neither the Act nor the Rules define the categories of “low-risk”, “non-low-risk”, or “high-risk” applicants. Specifically, Section 54 of the Act read with Rules 89 to 96A of the Rules, exhaustively set out that there would lie be a claim of refund provided there has been compliance with statutory conditions such as filing of returns, substantiation of zero-rated supplies, and absence of specific statutory disqualifications. By contrast, the impugned arrangement makes the timing and the availability of refunds contingent upon undisclosed algorithmic risk assessments, thereby introducing an extra-statutory eligibility filter. As a result, refundable ITC remains blocked not on the basis of any statutory infirmity, but due to opaque system-generated classifications that have no legislative provenance.
More fundamentally, the Instruction inverts the statutory architecture of Section 54(6) read with Rule 91. Under this framework, provisional refund is the rule, and withholding is the exception. Section 54(6) mandates the grant of provisional refund (up to 90%) in cases of zero-rated supplies, subject only to the limited carve-outs recognised in Rule 91, which require the proper officer to form a case-specific, reasoned opinion, even where risk considerations are involved. The power to withhold is thus individualised, discretionary, and accompanied by procedural safeguards. The Instruction, however, mandates a blanket denial of provisional refunds to all applicants not categorised as “low-risk” by the system, regardless of individual compliance or factual context. In doing so, it reverses the statutory presumption in favour of early refund, substitutes legislative standards with vague risk labels, and mechanically replaces the discretion of the proper officer granted by the Act (even though the arbitrary exercise of such power also is undesirable), reducing a statutory decision to a system-driven outcome.
‘Impact’ due to the ‘excessive delegation’
Since, the overall ‘purpose’ of Section 54(6) contemplates timely provisional refunds subject to later checks, accordingly, a high‑level, automated risk classification if made a precondition to release of refunds, would distort the statutory design and undermine the natural expectation of prompt refunds.
In this realm, as settled by the SC, delegated legislation cannot add ‘substantive conditions’ beyond those expressly provided in the parent Act. It may be submitted that the introduction of ‘high-low-risk’ criteria amounts to an ‘impermissible delegation’ of substantive decision-making vis-à-vis statutory right of refund to algorithmic parameters which have not been contemplated in the parent statute. The Act does not confer upon the CBIC an open-ended power to condition, curtail, or suspend statutorily guaranteed refunds, on the basis of internal risk ratings. Admittedly, while the Act permits administrative facilitation and procedural regulation, it does not authorise the executive to effectively amend the refund entitlement itself. The Instruction thus operates not as a procedural aid, but as an executive rule that alters Section 54, in substance, without any clearly defined criteria or safeguards.
It is also well settled that subordinate legislation and executive instructions cannot override, supplant, or dilute statutory rights. In Indian Express, the SC affirmed that delegated legislation is subject to judicial scrutiny and may be struck down where it exceeds the ‘scope’ of the enabling statute or is manifestly arbitrary. Similarly, in Cellular Operators, the court held that regulatory action must be based in the parent statute and cannot travel beyond the contours of delegated authority. Applying these principles, the impugned Instruction is vulnerable on both counts (i) it lacks statutory backing and (ii) defeats the legislative objective of early liquidity release, which lies at the core of the GST refund framework.
Provisional refund as exception rather than norm in practice
Textually, Section 54(6) r/w Rule 91 continues to treat provisional refund as the ideal position because of the use of imperative language (“shall grant”) and the legislative intent of ensuring liquidity for businesses, especially MSME’s and exporters.
However, the impugned arrangement mandates a system-generated “non-low-risk” classification as a de facto categorical bar to provisional refund. This reverses the statutory logic by shifting the officer’s role from exercising independent judgment to mechanically implementing a system outcome, thereby converting an exception into a default exclusion.
In light of the above, one might maintain that the impugned classification, because it withholds GST refunds from purportedly “high‑risk” applicants, also violates the canon of equity. This is because such an imposition-classification is unrelated to ‘ability to pay’ or income and it arbitrarily shifts a heavier, regressive liquidity burden onto purportedly “high‑risk” applicant-assessee, which are often MSMEs and exporters. Notably, it breaches the canon of certainty as well, because it makes the timing, manner and realisation of a refund unpredictable while being contingent on administrative discretion rather than clear ‘law’. which has already been determined by Rule 89(5) formula. This arguably is a clear evidence of ‘how’ both taxpayer liability and cash‑flow consequences are made (unnecessarily) unpredictable.
Constitutional implications vis-a-vis Article 19(1)(g) read with Article 19(6)
Even after the amendment to Rule 91(2) of the Rules (which incorporates system‑based risk scoring into the provisional refund process), the regime remains constitutionally vulnerable because it permits prolonged retention of refundable input tax credit (ITC) based on undisclosed risk classifications rather than clear statutory standards.
Under Article 19(1)(g), every citizen has the ‘freedom to carry on any trade or business’. The SC has repeatedly held that economic regulation which renders business commercially impracticable or imposes undue constraints on capital flow amounts to an infringement of this freedom.
Observably, since delays in refunding ITC, especially when provisional refund (90 %) is automatically withheld from “non‑low‑risk” applicants, have a direct liquidity impact on manufacturers, MSMEs and exporters, who rely on refund realisation to fund working capital. Accordingly, per Saghir Ahmad and considering the practical reality that when such delays accumulate over time, the refund entitlement arguably loses its commercial utility and then effectively acts as a restriction on business operations.
The risk framework, if treated as binding, effectively prescribes a substantive law by administrative means, contrary to the constitutional principle that taxation and refunds are governed by statute and law. To be a valid restriction under Article 19(6), the impugned measure must be (a) imposed by law, and (b) reasonable in the interests of the general public. Here, even assuming the impugned arrangement were to be a law, the measure is not ‘reasonable’, it being not “in the interests of the general public”. This is primarily because less restrictive alternatives, such as post‑refund audit, retention of amounts as a conditional lien, or recovery proceedings under Sections 73-74 where irregularities are later discovered, could protect revenue without withholding refunds for prolonged periods from compliant taxpayers.
The risk‑based 90% provisional GST refund mechanism fails the proportionality test under Article 19(6) because it imposes a broad, opaque, and punitive withholding of refunds on the basis of an algorithmic risk tag with no disclosed criteria, thresholds, or opportunity for contestation. There is no demonstrable direct nexus shown between the risk classification and the public interest; less restrictive measures (e.g., post‑refund audits, targeted verification, or assurances such as bank guarantees) could achieve revenue protection without crippling liquidity for compliant taxpayers. This violates the proportionality stricto sensu and the rational nexus components identified in Modern Dental, which require a ‘proper purpose’, a ‘rational connection’, ‘necessity’, and ‘proportionality’ between the measure and its objective.
Since taxpayers and exporters structure their affairs on the basis of the statutory refund regime and its timelines; a sudden, opaque risk classification that alters refunds undermines legitimate expectations and the predictability essential to business. Therefore, this measure also falls foul per the ‘doctrine of legitimate expectations’, because there are no legally ascertainable standards of risk classification that taxation subjects must expect or foresee and conform to.
Conclusion
This piece has examined the ‘impact’ and ‘disproportion’ of the CBIC’s imposition of a ‘risk-based’ criteria for refund of tax, namely ITC. This piece has argued that the ‘risk-based provisional refund’ regime under GST is legally unsustainable on three grounds. Firstly, despite the amendment to Rule 91(2), it is ultra vires Section 54 by making refunds contingent on undefined system-based classifications. Secondly, the Instruction reflects excessive delegation, converting a statutory entitlement into an administrative privilege. Thirdly, there lies a constitutional reason to reconsider this arrangement as it is unreasonable, giving no chance of ‘fair hearing’ to the concerned taxpayer-applicant and in not considering ‘lesser restrictive measures’, which are validly available, curtails the Article 19(1)(g) freedom of ‘trade, occupation and business’ of MSMEs and exporters, by blocking the working capital which ought be legitimately at their disposal but for the ‘high-risk’-based withholding.
In conclusion, even if a limited risk exercise is permissible as a matter of policy, a blanket conditionality on 90% provisional refunds is disproportionate. Therefore, a more targeted relocation of risk management to the post-refund stage through audit and recovery, rather than employing it to obstruct statutorily vested input-tax-credit refunds. This is essentially because so long as risk classification remains opaque, non-reviewable, and unbounded in its ‘scope’ and ‘timeline’, its deployment as a ground for refund denial is administratively distortive, economically punitive and constitutionally suspect, particularly for MSMEs and exporters.
The analysis therefore, in principle, calls for re-consideration of the risk-based standard to specifically define the parameters on which risk is to be evaluated, publicly. Alternatively, there the court may maintain enforceable limits on CBIC’s (and concerned officer’s) discretion with strictly time-bound refund mechanisms to realign GST administration with principles of natural justice and fair play.
Comments